Origin of Ostryopsis Intermedia (Betulaceae) in the Southeast Qinghaitibet Plateau Through Hybrid Speciation

Journal of Systematics and Evolution 52 (3): 250–259 (2014) doi: 10.1111/jse.12091

Research Article Origin of Ostryopsis intermedia (Betulaceae) in the southeast Qinghai–Tibet Plateau through hybrid speciation 1Zhi‐Qiang LU† 2Bin TIAN† 1Bing‐Bing LIU 1Chen YANG 1Jian‐Quan LIU* 1(State Key Laboratory of Grassland Agro‐Ecosystem, School of Life Sciences, Lanzhou University, Lanzhou 730000, China) 2(Key Laboratory of Biodiversity Conservation in Southwest China, State Forestry Administration, Southwest Forestry University, Kunming 650224, China)

Abstract It remains unknown how Ostryopsis intermedia B. Tian & J. Q. Liu (Betulaceae) originated in the SE Qinghai–Tibet Plateau (QTP). We examined sequence variations of two maternally inherited chloroplast (cp) DNA fragments and ampliﬁed fragment length polymorphisms (AFLP) of 32 populations for this species and two congeners, O. davidiana Decne. and O. nobilis Balf. f. W. W. Sm., distributed in northern China and the SE QTP. Each of the two distinct cpDNA haplotype groups was ﬁxed by O. davidiana or O. intermedia, respectively, whereas both were found in O. nobilis. All analyses of AFLP suggested that O. intermedia is more closely related to O. davidiana than to O. nobilis and Bayesian clustering analysis (K ¼ 2) further suggested that the genomic composition of O. intermedia showed a slightly mosaic origin from the other two species, although mostly from O. davidiana. Analyses of AFLP indicated the highest nuclear genetic diversity for O. davidiana and lowest for O. intermedia. The origin of O. intermedia through hybrid speciation due to southward migration of O. davidiana and the following hybridization with O. nobilis may provide the most reasonable explanation for the inconsistency between phylogenetic relationships of three species and degrees of genetic diversity from cpDNA sequences or AFLP datasets. Ostryopsis intermedia may represent one more diploid hybrid species found in the QTP, although further strong tests are needed based on additional data. Key words diploid hybrid speciation, Ostryopsis intermedia, QTP.

Speciation usually involves the production of two et al., 1995; Buerkle et al., 2000). However, despite novel species from a common ancestor through being rare, more homoploid hybrid species have been geographical isolation and/or adaptive evolution medi- reported due to the availability of diverse molecular ated by natural selection (Coyne & Orr, 2004; Riese- markers for determining the genetic origins of potential berg & Willis, 2007). However, two ancestral species hybrid species (Rieseberg, 2000; Mallet, 2007; Brels- can “melt” into the third species through hybridization ford et al., 2011; Sun et al., 2014). with two ancestral species still retained (Soltis & The mountains of the Qinghai–Tibet Plateau Soltis, 2009; Nolte & Tautz, 2010). This is particularly (QTP) constitute one of the world’s biodiversity common in plants, especially through polyploidizations hotspots with more than 9000 vascular plants, of which (Barton & Hewitt, 1985, 1989; Arnold, 1992, 1997, 18% are endemic to this area (Wu, 2008). The rapid 2006; Gross et al., 2003; Gross & Rieseberg, 2005; and continuous uplifts of the QTP might have driven Abbott et al., 2013). However, homoploid hybrid radiative diversification for a few plant genera (e.g., Liu speciation is considered evolutionarily more difficult et al., 2006). In addition, the new niches created by the (Rieseberg, 1997; Buerkle et al., 2000; Gross & uplifts as well as those by plant retreats in response to Rieseberg, 2005; Nolte & Tautz, 2010). This has the climatic changes in this climate‐sensitive region been explained by the theoretical prediction that also provide strong opportunities to develop new hybrid reproductive isolations between homoploid hybrids species. Polyploid hybrid speciation seems to have and their parents are difficult to achieve (McCarthy occurred less frequently than previously thought during plant diversification in this region (Liu, 2004). Howev- er, only two diploid hybrid gymnosperm species were reported from this region (Ma et al., 2006; Sun Received: 24 December 2013 Accepted: 12 March 2014 et al., 2014). In this study, we aim to examine whether † These authors contributed equally to this work. Author for correspondence. E‐mail: [email protected]. Tel.: 86‐931‐ an angiosperm species, Ostryopsis intermedia B. Tian 8914305. Fax: 86‐931‐8914288. & J. Q. Liu may represent a third diploid hybrid species

© 2014 Institute of Botany, Chinese Academy of Sciences LU et al.: Diploid hybrid origin of Ostryopsis intermedia (Betulaceae) 251 from the QTP. The genus Ostryopsis, consisting of using an eTrex handheld GPS unit (Garmin, Taiwan, three diploid species (2n ¼ 16), is an endemic genus in China). China (Li & Skvortsov, 1999; Tian et al., 2010). Both O. intermedia and O. nobilis Balf. f. W. W. Sm., are 1.2 DNA extraction, amplification, plastid DNA distributed in the SE QTP whereas O. davidiana Decne. sequencing, and AFLP fingerprinting occurs in northern China with distinct disjunction. Total genomic DNA was extracted from approxi- Although O. intermedia and O. nobilis overlap in one mately 20–30 mg silica gel‐dried leaf according to a site, the remaining distributions of two species are modified CTAB procedure (Doyle & Doyle, 1987). most allopatric with O. intermedia distributed more Several pairs of cpDNA primers designed by Hamilton southward. However, O. intermedia shows a morpho- (1999), Taberlet et al. (1991), and Sang et al. (1997) logical combination of the other two species (Fig. S1). were used in the initial screening, in which four or five Further phylogenetic constructions based on nuclear individuals were screened from each population. Two internal transcribed spacer (ITS) sequence variation pairs of primers, trnL‐trnF and psbA‐trnH (Hamilton, suggested that O. intermedia is closely related to O. 1999), which were identified to have sequence davidiana, rather than to the adjacent O. nobilis. All variations between the sampled individuals, were these findings suggested that the origin of O. intermedia therefore used for all remaining individuals. We carried is rather complex and it may represent a new diploid out polymerase chain reaction (PCR) in a 25‐mL hybrid species if evaluated from morphological volume, which contained 10–40 ng plant DNA, combinations. 50 mmol/L Tris–HCl, 1.5 mmol/L MgCl2, 0.5 mmol/L In this study, we collected typical populations of dNTPs, 2 mmol/L each primer, and 0.75 U Taq the three species across their distributions. We further polymerase. The program used for DNA amplification used nuclear data (amplified fragment length poly- in a T1 thermocycler (Biometra, Göttingen, Germany) morphisms [AFLP] fingerprinting) to confirm interspe- was coded for an initial denaturation step at 94 °C for cific relationships previously revealed by ITS at the 6 min, followed by 36 cycles of 45 s at 94 °C, 50 s at population level and examine the genetic diversity of 54 °C (psbA‐trnH)or56°C(trnL‐trnF), 90 s at 72 °C, each species. At the same time, we also examined and a final 7‐min extension step at 72 °C. All PCR interspecific relationships and genetic diversity of the products were then purified using a TIANquick Midi three species based on chloroplast (cp) DNA sequence Purification Kit following the manufacturer’s protocol variations. In contrast to the biparental inheritance (Tiangen, Beijing, China). Sequencing reactions were reflected by nuclear DNA (AFLP), cpDNA is usually carried out with the PCR primers, to cover the whole maternally inherited through seeds (Corriveau & PCR segment, using an ABI Prism BigDye terminator Coleman, 1988; Birky, 1995; Mogensen, 1996). The cycle sequencing ready reaction kit (Applied Biosys- inconsistent interspecific relationships at the molecular tems, Foster City, CA, USA). The reaction mixtures markers with contrasted inheritance backgrounds were analyzed on an Applied Biosystems model 3130xl usually provide strong evidence for interspecific automated sequencer (Applied Biosystems). Both trnL‐ introgression and hybrid speciation (e.g., trnF and psbA‐trnH sequences were obtained for 472 Arnold, 1997; Gross et al., 2003; Song et al., 2003). individuals of 32 populations of three species. We used a slightly modified protocol (Vos et al., 1995) to examine the AFLP of all collected 1 Material and methods populations. To test the reproducibility of AFLP fragments and to allow an estimation of the error 1.1 Sample collection rate, 20 individuals from 15 populations were replicated We sampled 32 populations of Ostryopsis davidi- from the restriction–ligation step onwards. An initial ana, O. nobilis, and O. intermedia across their screening of selective primers using 15 primer geographical ranges (Table 1). At only one locality, combinations was carried out. We selected three both O. nobilis and O. intermedia occur together while EcoRI/MseI primer combinations producing the most at the others, only one species was found (Fig. 1: A; repeatable and unambiguously scorable profiles: Table 1). Fresh leaves, spaced at least 100 m apart, EcoRI‐ACA/MseI‐CAT, EcoRI‐ACC/MseI‐CTG, and were collected from each population of each species EcoRI‐AGG/MseI‐CTT. Fluorescence‐labeled frag- and dried immediately in silica gel. Representative ments were separated on a CEQ 8000 capillary vouchers were deposited in the Lanzhou University sequencer (Beckman Coulter, Brea, CA, USA), with herbarium (Lanzhou, China). The latitude, longitude, an internal size standard. All loci of approximately and elevation of each collection location were measured 500 bp for all individuals were then visually inspected

Table 1 Location of populations, number of individuals sampled, and chloroplast DNA haplotype distribution per population of three species of Ostryopsis, and genetic diversity and variation statistics detected by ampliﬁed fragment length polymorphism (AFLP)

Population Location Latitude Longitude Altitude N1 Haplotypes AFLP (all in China) (N) (E) (m a.s.l.) (No. of individuals) N2 PLP (%) HE O. davidiana 1 Lixian, Sichuan 31°250 103°070 1950 8 H6 (8) 5 7.1 0.0703 2 Maoxian, Sichuan 31°390 103°480 1570 15 H8 (15) 10 68.8 0.171 3 Zhouqu, Gansu 33°520 104°090 1530 8 H7 (8) 5 66.2 0.152 4 Tangchang, Gansu 33°590 104°280 1680 13 H6 (13) 6 14.9 0.122 5 Tianshui, Gansu 34°250 105°580 1275 13 H5 (13) 10 21.4 0.115 6 Xunhua, Qinghai 35°490 102°410 1990 18 H1 (18) 13 59.7 0.152 7 Yuzhong, Gansu 35°470 104°020 2450 14 H1 (14) 10 62.3 0.131 8 Pingliang, Gansu 35°330 106°330 1462 18 H1 (18) 5 14.3 0.103 9 Yijun, Shaanxi 35°280 109°090 1149 14 H4 (14) 0 –– 10 Yan’an, Shaanxi 36°340 109°250 1010 13 H4 (13) 6 5.2 0.0436 11 Yinchuan, Ningxia 38°440 105°550 2250 9 H4 (9) 5 11.0 0.0902 12 Qingxian, Shanxi 36°410 112°320 1070 16 H4 (16) 0 –– 13 Wuxiang, Shanxi 37°030 112°320 1340 12 H4 (12) 7 64.3 0.149 14 Wutai, Shanxi 38°570 113°290 1670 19 H4 (19) 0 –– 15 Huhehaote, IMG 40°540 111°340 1280 17 H3 (17) 6 13.6 0.0888 16 Lingqiu, Shanxi 39°070 113°570 920 13 H2 (13) 6 9.1 0.0759 17 Zhuolu, Hebei 40°120 115°090 940 11 H2 (11) 10 15.6 0.0959 18 Linyuan, Liaoning 40°550 119°160 720 18 H2 (18) 10 13.0 0.0861 19 Xiaowutai, Hebei 39°580 115°040 1630 17 H1 (11), H4 (6) 10 63.0 0.1426 20 Reshui, IMG 43°270 117°430 1100 15 H1 (15) 10 9.7 0.0636 21 Tongliao, IMG 44°570 120°260 650 14 H1 (14) 10 18.2 0.103 Subtotal 295 144 46.8 O. nobilis 22 Daju, Yunnan 27°160 100°130 1910 15 H9 (15) 13 11.7 0.0674 23 Muli, Sichuan 27°490 101°120 2190 19 H14 (19) 10 11.0 0.0609 24 Jiren, Yunnan 27°480 99°280 1950 18 H11 (18) 11 11.0 0.0753 25 Tangdui, Yunnan 27°590 99°340 2900 20 H12 (10), H13 (10) 17 14.9 0.0821 26 Nixi, Yunnan 28°090 99°270 2540 12 H12 (2), H10 (10) 15 15.6 0.0821 Subtotal 84 66 22.1 O. intermedia 27 Deqin, Yunnan 28°220 98°540 2870 20 H12 (20) 6 9.7 0.07605 28 Weixi, Yunnan 27°130 98°590 1880 16 H12 (16) 10 14.3 0.0657 29 Badi, Yunnan 27°290 98°010 1780 12 H12 (12) 12 5.2 0.0372 30 Yezhi, Yunnan 27°350 99°020 1730 13 H12 (13) 12 6.5 0.0424 31 Jiren, Yunnan 27°480 99°280 1950 12 H12 (12) 0 –– 32 Judian, Yunnan 27°170 99°390 1950 20 H12 (20) 12 14.9 0.0932 Subtotal 93 52 17.5 Total 472 262

–, No calculation. HE, Nei’s gene diversity; IMG, Inner Mongolia; N1, number of trees analyzed for cpDNA data; N2, number of trees analyzed for AFLP data; PLP, proportion of polymorphic loci. and manually adjusted. Only unambiguously scorable 3.11 (Excoffier et al., 2005). Interspecific relationships loci and individuals were included in the analysis and were constructed among cpDNA haplotypes, using peaks found in <3% of individuals were excluded. In median‐joining networks with Network version 4.6.1.1 total, 262 individuals were scored for 154 markers. To (Bandelt et al., 1999; available at http://www.fluxus‐ ensure high repeatability of analyzed AFLP loci, we ran engineering.com). Haplotype diversity (Hd) and nucle- a subset of 30 individuals twice from the preselective otide diversity (p) were calculated in DnaSP version amplification step. 5.10.01 (Librado & Rozas, 2009). Within‐population diversity (HS) and total diversity (HT) were calculated 1.3 Data analysis using the program PERMUT (Pons & Petit, 1996). All cpDNA sequences were aligned and edited Based on AFLP, we assessed the genetic diversity with MEGA version 5.0 (Tamura et al., 2011). Both psbA‐ of each species: the percentage of loci that are trnH and trnL‐trnF sequences were deposited in the polymorphic, Nei’s expected heterozygosity (HE), total GenBank database under accession numbers diversity (HT), within‐population diversity (HS), be- EU852623 to EU852631. Indels longer than 1 bp tween‐population diversity (HB), and population dif- were all treated as single mutation events. Pairwise FST ferentiation (FST) using the method of Lynch & between species was calculated with Arlequin version Milligan (1994) as implemented in the program AFLP‐

Fig. 1. Distributions and networks of haplotypes within three species of Ostryopsis. A, Geographical map of sampling locations (all in China) and distribution of haplotypes for chloroplast DNA. B, Network of 14 haplotypes. For each haplotype, circle size is proportional to its frequency over all populations. Population codes (locality, province) for O. davidiana: 1, Lixian, Sichuan; 2, Maoxian, Sichuan; 3, Zhouqu, Gansu; 4, Tangchang, Gansu; 5, Tianshui, Gansu; 6, Xunhua, Qinghai; 7, Yuzhong, Gansu; 8, Pingliang, Gansu; 9, Yijun, Shaanxi; 10, Yan’an, Shaanxi; 11, Yinchuan, Ningxia; 12, Qingxian, Shanxi; 13, Wuxiang, Shanxi; 14, Wutai, Shanxi; 15, Huhehaote, Inner Mongolia (IMG); 16, Lingqiu, Shanxi; 17, Zhuolu, Hebei; 18, Linyuan, Liaoning; 19, Xiaowutai, Hebei; 20, Reshui, IMG; 21, Tongliao, IMG. Population codes for O. nobilis: 22, Daju, Yunnan; 23, Muli, Sichuan; 24, Jiren, Yunnan; 25, Tangdui, Yunnan; 26, Nixi, Yunnan. Population codes for O. intermedia: 27, Deqin, Yunnan; 28, Weixi, Yunnan; 29, Badi, Yunnan; 30, Yezhi, Yunnan; 31, Jiren, Yunnan; 32, Judian, Yunnan.

SURV version 1.0 (Vekemans et al., 2002) with the 2 Results default model (Zhivotovsky, 1999). To reveal interspecific differentiation we calculated pairwise Nei’s 2.1 Chloroplast DNA analysis genetic distances and FST values between species using From the 472 individuals sampled, we detected AFLP‐SURV version 1.0 (Vekemans et al., 2002). The two and eight nucleotide substitutions for the trnL‐trnF significance of FST values was calculated based on and psbA‐trnH fragments, respectively. The aligned comparison with values obtained from 10 000 randomly sequences of the psbA‐trnH spacer were 445 bp in permuted individuals between species. To assess length, and the length of the aligned trnL‐trnF DNA genetic structure, we carried out a principal coordinate sequences is 822 bp (including trnL and the trnL‐trnF analysis (PCA) based on a genetic distance matrix spacer region) with four insertions (Table S1). All generated from GenAlEx version 6.2 (Peakall & variations across both fragments (1267 bp in length) Smouse, 2006). We also ran Bayesian clustering identified 14 haplotypes (H1–H14) (Tables 1, S1). All analyses using STRUCTURE version 2.3.3 (Hubisz individuals from Ostryopsis intermedia shared one et al., 2009). Simulations were carried out based on haplotype (H12) and this haplotype was also found in the admixture model with correlated allele frequencies O. nobilis. The other five haplotypes were further (Bonin et al., 2007). We carried out 20 runs with a burn‐ recovered from O. nobilis whereas eight haplotypes in of 500 000 generations and Markov chains of were identified for all exampled individuals from 800 000 generations (1 K 8). The most likely O. davidiana. All haplotypes from three species clus- number of clusters was estimated using the method tered into two clades (Fig. 1: B). One clade consisted of originally described by Pritchard et al. (2000) and also the haplotypes from O. nobilis (H9–H12) and the DK approach described by Evanno et al. (2005). The O. intermedia (H12) while the other comprised those neighbor‐joining tree was calculated with FAMD recovered from O. davidiana (H1–H8) and O. nobilis version 1.30 (Schlüter & Harris, 2006). (H13 and H14) (Fig. 1). In the only location where two

© 2014 Institute of Botany, Chinese Academy of Sciences 254 Journal of Systematics and Evolution Vol. 52 No. 3 2014 species co‐occurred, two different haplotypes were frequency >0.9 in a particular species and <0.05 in found, respectively, for each species, O. intermedia other species. The number of private bands in each (H12) and O. nobilis (H13) (Fig. 1). species was 6 (O. davidiana), 10 (O. nobilis), and 5 Nucleotide diversity (p), haplotype diversity (Hd), (O. intermedia). The mean number of AFLP bands of and total genetic diversity of O. nobilis (p ¼ 0.00217, O. intermedia (79.9) (per individual per species) was Hd ¼ 0.832, and HT ¼ 0.992) were higher than those of slightly higher than those in O. davidiana (79.3) and O. davidiana (p ¼ 0.00186, Hd ¼ 0.785, and HT ¼ O. nobilis (73.5). However, when the five private bands 0.836), whereas all these estimates were lowest in were excluded, the average band number of O. intermedia (Table 2). Estimates of genetic differen- O. intermedia (75.9) was intermediate between those tiation between O. intermedia and O. davidiana (FST ¼ of the other two species (Fig. 2). 0.7089) in terms of pairwise FST were higher than The estimates of the genetic diversity for each the differentiations between O. davidiana and species are listed in Table 1. Among the three species, O. nobilis (FST ¼ 0.4274) and between O. nobilis and the total population diversity (HT), average within‐ O. intermedia (FST ¼ 0.4102) (Table S2). population diversity (HS), and average between‐ population diversity (HB)ofO. davidiana (HT ¼ 2.2 Analysis of AFLP 0.1906; HS ¼ 0.1087; HB ¼ 0.0819) were higher than The average per‐locus genotyping error rate for the O. nobilis (HT ¼ 0.0934; HS ¼ 0.0736; HB ¼ 0.0198) AFLP data, measured as recommended by Bonin et al. and O. intermedia (HT ¼ 0.0750; HS ¼ 0.0629; HB ¼ (2004), is 2.55%. Of a total of 154 bands surveyed, 127 0.0121). In addition, the overall diversity of O. nobilis (82.5%) were polymorphic. Private bands were defined was also higher than that of O. intermedia (Table 3). as the presence or absence of an AFLP band at The interspecific FST differentiations between O. intermedia and O. davidiana (FST ¼ 0.5622, Nei’s genetic distance ¼ 0.1960) were clearly lower than Table 2 Genetic diversity estimates for chloroplast DNA in three those between O. intermedia and O. nobilis (FST ¼ species of Ostryopsis 0.724, Nei’s genetic distance ¼ 0.2580) and those p Species Hd HT HS between O. davidiana and O. nobilis (FST ¼ 0.5861, O. davidiana 0.00186 0.785 0.836 0.023 Nei’s genetic distance ¼ 0.2356) (Table S3). O. nobilis 0.00217 0.832 0.992 0.166 Using the method originally described by Pritchard O. intermedia –––– Total 0.00277 0.862 0.892 0.041 et al. (2000) and also the DK approach described by

–, No calculation. p, nucleotide diversity; Hd, haplotype diversity; HS, Evanno et al. (2005), we found the most likely number average gene diversity within populations; HT, total gene diversity. of Bayesian clusters was three (K ¼ 3) (Fig. S2). When

Fig. 2. Mean number of ampliﬁed fragment length polymorphism (AFLP) bands per individual per Ostryopsis species. A, Five private alleles in O. intermedia are excluded. B, No band is excluded.

Table 3 Genetic diversity statistics for amplified fragment length between O. davidiana and O. nobilis but closer to polymorphism in three species of Ostryopsis O. davidiana (Fig. 4). The neighbor‐joining analyses Species HT HS HB FST recovered three distinct and highly supported clades O. davidiana 0.1906 0.1087 0.0819 0.4302 (bootstrap value ¼ 100), corresponding to the three O. nobilis 0.0934 0.0736 0.0198 0.2142 O. intermedia 0.0750 0.0629 0.0121 0.1595 species (Fig. 5). Total 0.2502 0.0942 0.1559 0.6232 Significant at P < 0.001. FST, population differentiation; HB, average ‐ ‐ between population diversity; HS, average within population diversity; 3 Discussion HT, total population diversity. In order to explore the possible origin pattern of Ostryopsis intermedia, we further examined genetic K ¼ 2, individuals from O. davidiana and O. nobilis differentiation and interspecific relationships between clustered to different groups with high probability, this species and the other two congeners based on two whereas individuals from O. intermedia were assigned sets of molecular markers with contrasting inheriting to both groups with slightly mosaic assemblies, and background at the population level. Our analyses of the more closely to O. davidiana (Fig. 3). When K ¼ 3, maternally inherited cpDNA sequences from all individuals from O. davidiana and O. nobilis were still populations of three species suggested that O. inter- assigned to their respective clusters, but individuals media shares a haplotype with the nearby O. nobilis, but from O. intermedia were assigned to a distinct, third with a distinct relationship with O. davidiana. The cluster (Fig. 3). The PCA based on a genetic distance genetic diversity of O. nobilis is higher than those of the matrix divided the data set into three clusters. other two species. However, based on biparentally Dimension 2 (26.69%) separated O. intermedia from inherited AFLP markers, we found that O. intermedia is the other two species, and in Dimension 1 (58.81%), more closely related to the distantly distributed individuals from O. intermedia were intermediate O. davidiana than to O. nobilis. In addition, the genetic

Fig. 3. Structure analysis of Ostryopsis davidiana, O. nobilis, and O. intermedia when K ¼ 2 and K ¼ 3 clusters are assumed. Each vertical line represents an individual, with each color corresponding to the posterior probability of assignment to each of a number of clusters (K).

Fig. 4. Principal coordinate analysis of ampliﬁed fragment length polymorphism marker variation among 262 individuals in Ostryopsis.

diversity of O. davidiana based on AFLP is higher than recovered from O. intermedia and O. davidiana were that of O. nobilis. Interspecific introgressions and placed within separate clades, indicating their distant hybrid speciation seem to be the most parsimonious relationship. Nonetheless, six haplotypes of O. nobilis explanation for these inconsistences between interspe- were distributed in both clades. In addition, the genetic cific relationships constructed by two sets of molecular diversity of O. nobilis is higher than that of the other two markers. species. These findings suggested two alternative evolutionary scenarios occurring between the three 3.1 Inconsistent interspecific relationships be- species (Arnold, 1997). First, O. nobilis is an ancestral tween two sets of molecular markers species, from which two daughter species, O. inter- All haplotypes based on cpDNA sequence media and O. davidiana, originated, and therefore variations clustered into two clades with two distinct comprised only one of two ancestral lineages. This steps (Fig. 1). It is interesting that the haplotypes hypothesis is also consistent with the high cpDNA diversity found for O. nobilis. Second, O. intermedia is closely related to O. nobilis from the maternally inherited cpDNA, and they both diverged from O. davidiana earlier. However, cpDNA introgressions from O. davidiana were fixed in O. nobilis. This is supported by the finding that two haplotypes (H13 and H14) of O. nobilis clustered with those of O. davidiana in a separate clade seem to originate from the most common haplotype found in the latter species (Tian et al., 2009). Interestingly, AFLP analyses suggested different interspecific relationships. Based on PCA and neighbor‐joining analyses of all sampled individuals from three species, O. intermedia is closely related to O. davidiana and they together diverged from O. nobilis at an early stage (Figs. 4, 5). This is totally consistent with the early interspecific relationships constructed based on the nuclear ITS sequence variation (Tian et al., 2009). This interspecific relationship failed to support the hypothesis that O. nobilis is the ancestor species to both O. intermedia and O. davidiana.In Fig. 5. Unrooted neighbor‐joining tree for three species of Ostryopsis. addition, the genetic diversity of O. nobilis is not

© 2014 Institute of Botany, Chinese Academy of Sciences LU et al.: Diploid hybrid origin of Ostryopsis intermedia (Betulaceae) 257 higher than the other two “daughter species” as suggested its rapid expansion possibly after its hybrid expected from the “progenitor–derivative” speciation origin. (Tables 1, 3). However, we found that the genetic Because of climatic oscillations, most plant species diversity of O. davidiana is the highest based on AFLP distributed in northern China retreated southward during (Table 3). This agrees well with the fact that the the glacial stages (Tian et al., 2009; Qiu et al., 2011; distribution of this species is wider than those of the Sakaguchi et al., 2012). Therefore, it is highly likely that other two congeners (Ma et al., 2006; Sun et al., 2014). O. davidiana retreated southward, contacted and Therefore, the haplotypes gathering together with those hybridized with O. nobilis, which finally resulted in from O. davidiana found in O. nobilis is more likely to the cpDNA introgressions from O. davidiana to derive from the introgressions from the early diverged O. nobilis. From these hybrid offspring, a new diploid O. davidiana. lineage, O. intermedia, arose, after initial competition with both parents. Possibly due to the retreat of 3.2 Introgression and hybrid speciation O. davidiana back to northern China again, competition In addition to the possible cpDNA introgression between this new species and O. davidiana was from O. davidiana to O. nobilis as inferred through weakened. In addition, as suggested by AFLP and ITS the above analyses, the inconsistent interspecific data, this hybrid species inherited most nuclear elements relationship of O. intermedia with the other two from O. davidiana, therefore it is very easy for it to species recovered from two different sets of molecular develop reproductive isolation from the parapatrically markers strongly indicates its origin through hybrid distributed O. nobilis. However, it also inherited fewer speciation. Both AFLP and ITS data supported the nuclear elements from O. nobilis in addition to the close relationship between distantly disjunct cpDNA elements, as mentioned before. Furthermore, O. intermedia and O. davidiana. If the nuclear O. intermedia may have expanded its distributional interspecific relationship is assumed to be true, the range with a severe founder effect in the new niche shared cpDNA haplotype between O. intermedia and different from the O. nobilis, which resulted in the O. nobilis has to be hypothesized to be introgressed fixation of the single haplotype across its current from the latter species to the former. However, this distributional range. hypothesis did not receive further support from the Two well‐characterized diploid hybrid species, haplotype distribution of both species. The introgres- Pinus densata and Picea purpurea, were reported from sions usually occurred in the parapatric distributions the QTP (Mao & Wang, 2011; Wang et al., 2011b; Gao of two species and replaced only a part of the alleles et al., 2012; Sun et al., 2014). Both conifer species or haplotypes in the introgressed species (e.g., occupied the habitats from which their parents were McGuire et al., 2007; Du et al., 2011; Wang et al., excluded. Our study seems to indicate that O. intermedia 2011a). However, we found that O. intermedia and represents a third diploid hybrid species, with most of its O. nobilis were found to fix different haplotypes in the distributions also different fromthe two potential parents. only co‐occurring site. In addition, all populations of Therefore, the new habitats created by the QTP uplifts O. intermedia were fixed by one haplotype. Origin and the following climatic oscillations might have of O. intermedia by hybrid speciation provides facilitated production of the homoploid hybrid species. the most parsimonious explanation for these contra- With the applications of the different molecular dictory findings. First, although nuclear compositions markers to population genetic studies of more species, of O. intermedia derived mostly from those of it is expected that more homoploid speciation species O. davidiana, this species still possesses a slightly will be reported from the QTP biodiversity hotspot. mosaic genome (K ¼ 2) with O. nobilis based on the AFLP (Figs. 2, 4). Second, O. intermedia is morpho- logically intermediate between the other two puta- Acknowledgements This study was supported by the tively parental species, as found for most diploid National High Technology Research and Development hybrid species (Fig. S1). Finally, most diploid species Program of China (863 Program, Grant No. were found to have experienced rapid expansion in 2013AA100605), the Research Fund for the Doctoral the unoccupied niche with decreased genetic diversity Program of Higher Education of China (Grant No. compared with the two parental species. Ostryopsis 20100211110008), the Collaboration Program from intermedia is mostly distributed out of the distribu- Ministry of Science and Technology of China (Grant tional ranges of the other two species and the low No. 2010DFB63500), and the Fundamental Research genetic diversity of this species at the cpDNA Funds for the Central Universities (Grant No. lzujbky‐ sequence variation with the single haplotype clearly 2009‐k05).

References Gao J, Wang B, Mao JF, Ingvarsson P, Zeng QY, Wang XR. 2012. Demography and speciation history of the homoploid Abbott R, Albach D, Ansell S, Arntzen JW, Baird SJ, Bierne N, hybrid pine Pinus densata on the Tibetan Plateau. Boughman J, Brelsford A, Buerkle CA, Buggs R, Butlin Molecular Ecology 21: 4811–4827. RK, Dieckmann U, Eroukhmanoff F, Grill A, Cahan SH, Gross BL, Rieseberg LH. 2005. The ecological genetics of Hermansen JS, Hewitt G, Hudson AG, Jiggins C, Jones J, homoploid hybrid speciation. Journal of Heredity 96: 241– Keller B, Marczewski T, Mallet J, Martinez‐Rodriguez P, 252. Möst M, Mullen S, Nichols R, Nolte AW, Parisod C, Gross BL, Schwarzbach RE, Rieseberg LH. 2003. Origins of the Pfennig K, Rice AM, Ritchie MG, Seifert B, Smadja CM, diploid hybrid species Helianthus deserticola (Asteraceae). Stelkens R, Szymura JM, Väinölä R, Wolf JB, Zinner D. American Journal of Botany 90: 1708–1719. 2013. Hybridization and speciation. Journal of Evolution- Hamilton MB. 1999. Four primer pairs for the amplification of – ary Biology 26: 229 246. chloroplast intergenic regions with intraspecific variation. Arnold ML. 1992. Natural hybridization as an evolutionary Molecular Ecology 8: 521–523. process. Annual Review of Ecology and Systematics 23: Hubisz MJ, Falush D, Stephens M, Pritchard JK. 2009. Inferring – 237 261. weak population structure with the assistance of sample Arnold ML. 1997. Natural hybridization and evolution. Oxford: group information. Molecular Ecology Resources 9: 1322– Oxford University Press. 1332. Arnold ML. 2006. Evolution through genetic exchange. Oxford: Li PC, Skvortsov AK. 1999. Betulaceae. In: Wu CY, Raven PH Oxford University Press. eds. Flora of China. Beijing: Science Press; St. Louis: ‐ Bandelt HJ, Forster P, Rohl A. 1999. Median joining networks Missouri Botanical Garden Press. 4: 284–313. fi for inferring intraspeci c phylogenies. Molecular Biology Librado P, Rozas J. 2009. DnaSP v5: A software for – and Evolution 16: 37 48. comprehensive analysis of DNA polymorphism data. Barton NH, Hewitt GM. 1985. Analysis of hybrid zones. Annual Bioinformatics 25: 1451. – Review of Ecology and Systematics 16: 113 148. Liu JQ. 2004. Uniformity of karyotypes in Ligularia Barton NH, Hewitt GM. 1989. Adaptation, speciation and hybrid (Asteraceae: Senecioneae), a highly diversified genus of – zones. Nature 341: 497 503. the eastern Qinghai‐Tibet Plateau highlands and adjacent Birky CW. 1995. Uniparental inheritance of mitochondrial and areas. Botanical Journal of the Linnean Society 144: 329– chloroplast genes: Mechanismsandevolution.Proceedingsof – 342. the National Academy of Sciences USA 92: 11331 11338. Liu JQ, Wang YJ, Wang AL, Hideaki O, Abbott RJ. 2006. Bonin A, Bellemain E, Bronken Eidesen P, Pompanon F, Radiation and diversification within the Ligularia‐Creman- Brochmann C, Taberlet P. 2004. How to track and assess thodium‐Parasenecio complex (Asteraceae) triggered by genotyping errors in population genetics studies. Molecular ‐ – uplift of the Qinghai Tibetan Plateau. Molecular Phyloge- Ecology 13: 3261 3273. netics and Evolution 38: 31–49. Bonin A, Ehrich D, Manel S. 2007. Statistical analysis of Lynch M, Milligan BG. 1994. Analysis of population genetic amplified fragment length polymorphism data: A toolbox structure with RAPD markers. Molecular Ecology 3: 91– for molecular ecologists and evolutionists. Molecular 99. Ecology 16: 3737–3758. Ma XF, Szmidt AE, Wang XR. 2006. Genetic structure and Brelsford A, Mila B, Irwin DE. 2011. Hybrid origin of evolutionary history of a diploid hybrid pine Pinus densata Audubon’s warbler. Molecular Ecology 20: 2380–2389. inferred from the nucleotide variation at seven gene loci. Buerkle CA, Morris RJ, Asmussen MA, Rieseberg LH. 2000. Molecular Biology and Evolution 23: 807–816. The likelihood of homoploid hybrid speciation. Heredity 84: 441–451. Mao JF, Wang XR. 2011. Distinct niche divergence character- Corriveau JL, Coleman AW. 1988. Rapid screening method to izes the homoploid hybrid speciation of Pinus densata on – detect potential biparental inheritance of plastid DNA and the Tibetan Plateau. American Naturalist 177: 424 439. – results for over 200 angiosperms. American Journal of Mallet J. 2007. Hybrid speciation. Nature 446: 279 283. Botany 75: 1443–1458. McCarthy EM, Asmussen MA, Anderson WW. 1995. A Coyne JA, Orr HA. 2004. Speciation. Sunderland: Sinauer theoretical assessment of recombinational speciation. Associates. Heredity 74: 502–509. Doyle JJ, Doyle JL. 1987. A rapid DNA isolation procedure for McGuire JA, Linkem CW, Koo MS, Hutchison DW, Lappin KA, small quantities of fresh leaf tissue. Phytochemical Bulletin Orange DI, Lemos‐Espinal J, Riddle BR, Jaeger J. 2007. 19: 11–15. Mitochondrial introgression and incomplete lineage sorting Du FK, Peng XL, Liu JQ, Lascoux M, Hu FS, Petit RJ. 2011. through space and time: Phylogenetics of crotaphytid Direction and extent of organelle DNA introgression lizards. Evolution 61: 2879–2897. between two spruce species in the Qinghai‐Tibetan Plateau. Mogensen HL. 1996. The hows and whys of cytoplasmic New Phytologist 192: 1024–1033. inheritance in seed plants. American Journal of Botany 83: Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of 383–404. clusters of individuals using the software STRUCTURE:A Nolte AW, Tautz D. 2010. Understanding the onset of hybrid simulation study. Molecular Ecology 14: 2611–2620. speciation. Trends in Genetics 26: 54–58. Excoffier L, Laval G, Schneider S. 2005. Arlequin (version 3.0): Peakall R, Smouse PE. 2006. GENALEX6: Genetic analysis in An integrated software package for population genetics data Excel. Population genetic software for teaching and analysis. Evolutionary Bioinformatics Online 1: 47–50. research. Molecular Ecology Notes 6: 288–295.

Pons O, Petit RJ. 1996. Measuring and testing genetic Tian B, Liu TL, Liu JQ. 2010. Ostryopsis intermedia, a new species differentiation with ordered versus unordered alleles. of Betulaceae from China. Botanical Studies 51: 257–262. Genetics 144: 1237–1245. Vekemans X, Beauwens T, Lemaire M, Roldán‐Ruíz I. 2002. Pritchard JK, Stephens M, Donnelly P. 2000. Inference of Data from amplified fragment length polymorphism population structure using multilocus genotype data. (AFLP) markers show indication of size homoplasy and Genetics 155: 945–959. of a relationship between degree of homoplasy and Qiu YX, Fu CX, Comes HP. 2011. Plant molecular phylogeog- fragment size. Molecular Ecology 11: 139–151. raphy in China and adjacent regions: Tracing the genetic Vos P, Hogers R, Bleeker M, Reijans M, Vandelee T, Hornes M, imprints of Quaternary climate and environmental change Fritjers A, Pot J, Peleman J, Kuiper M, Zabeau M. 1995. in the world’s most diverse temperate flora. Molecular AFLP: A new technique for DNA fingerprinting. Nucleic Phylogenetics and Evolution 59: 225–244. Acids Research 23: 4407–4414. Rieseberg LH. 1997. Hybrid origins of plant species. Annual Wang B, Mao JF, Gao J, Zhao W, Wang XR. 2011a. Review of Ecology and Systematics 28: 359–389. Colonization of the Tibetan Plateau by the homoploid Rieseberg LH. 2000. Crossing relationships among ancient and hybrid pine Pinus densata. Molecular Ecology 20: 3796– experimental sunflower hybrid lineages. Evolution 54: 3811. 859–865. Wang J, Abbot RJ, Peng YL, Du FK, Liu JQ. 2011b. Species Rieseberg LH, Willis JH. 2007. Plant speciation. Science 317: delimitation and biogeography of two fir species (Abies)in 910–914. central China: Cytoplasmic DNA variation. Heredity 107: Sakaguchi S, Qiu YX, Liu YH, Qi XS, Kim SH, Han J, Takeuchi 362–370. Y, Worth JRP, Yamasaki M, Sakurai S, Isagi Y. 2012. Wu YH. 2008. The vascular plants and their eco‐geographical Climate oscillation during the Quaternary associated with distribution of the Qinghai‐Tibetan Plateau. Beijing: landscape heterogeneity promoted allopatric lineage diver- Science Press. gence of a temperate tree Kalopanax septemlobus (Aral- Zhivotovsky LA. 1999. Estimating population structure in iaceae) in East Asia. Molecular Ecology 21: 3823– diploids with multilocus dominant DNA markers. Molecu- 3838. lar Ecology 8: 907–913. Sang T, Crawford DJ, Stuessy TF. 1997. Chloroplast DNA phylogeny, reticulate evolution and biogeography of Paeonia Supplementary Material (Paeoniaceae). American Journal of Botany 84: 1120– 1136. Schlüter PM, Harris SA. 2006. Analysis of multilocus The following supplementary material is available fingerprinting data sets containing missing data. Molecular online for this article at http://onlinelibrary.wiley.com/ Ecology 6: 569–572. doi/10.1111/jse.12091/suppinfo: Soltis DE, Soltis PS. 2009. The role of hybridization in plant speciation. Annual Review of Plant Biology 60: 561–588. Fig. S1. Photographs showing phenotypic differences Song BH, Wang XQ, Wang XR, Ding KY, Hong DY. 2003. in leaves of three species of Ostryopsis.A, Cytoplasmic composition in Pinus densata and population establishment of the diploid hybrid pine. Molecular O. davidiana.B,O. intermedia.C,O. nobilis. Ecology 12: 2995–3001. Sun YS, Abbott RJ, Li LL, Li L, Zou JB, Liu JQ. 2014. Fig. S2. Estimated number of clusters (K) obtained with Evolutionary history of Purple cone spruce (Picea STRUCTURE version 2.3.3, using the original method purpurea) in the Qinghai‐Tibet Plateau: Homoploid hybrid (LnPD) from Pritchard et al. (2000) and the DK origin and Pleistocene expansion. Molecular Ecology 23: statistics given in Evanno et al. (2005). 343–359. Taberlet P, Gielly L, Pautou G, Bouvet J. 1991. Universal primers for amplification of three non‐coding regions of Table S1. Variable sites of the aligned sequences of chloroplast DNA. Plant Molecular Biology 17: 1105–1109. chloroplast DNA fragments in 14 haplotypes of Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. Ostryopsis. 2011. MEGA5: Molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and Table S2. Pairwise FST between species for chloroplast maximum parsimony methods. Molecular Biology and DNA. Evolution 28: 2731–2739. Tian B, Liu RR, Wang LY, Qiu Q, Chen KM, Liu JQ. 2009. Table S3. Matrix of pairwise FST (below diagonal) and Phylogeographic analyses suggest that a deciduous species ’ (Ostryopsis davidiana Decne., Betulaceae) survived in Nei s genetic distances (above diagonal) between northern China during the Last Glacial Maximum. Journal species in Ostryopsis for amplified fragment length of Biogeography 36: 2148–2155. polymorphism.